ANALYSIS OF WEP PERFORMANCE ON MOBILE
DEVICES
Arnulfo Ochoa Indart
1
, Jesús Arturo Pérez Díaz
2
,
1 Informatic Graduate Program, ITESM Campus Cuernavaca, Paseo de la Reforma 182-A,
Cuernavaca, México
2 Electronic and Communications Department, ITESM Campus Cuernavaca, Paseo de la
Reforma 182-A, Cuernavaca, México
Keywords. Wireless Networks, 802.11b, WEP, Performance, Mobile Devices
Abstract. Mobile devices are becoming more popular
every day; they must
keep up with security implemented by desktop computers. This paper tries to
evaluate performance of data transmission with and without ciphering tech-
niques. WEP is not the best way of securing a network but it is widely used,
that is why we used WEP on these tests. This article tries to define how much
performance is lost with WEP, so we can estimate the loss of performance on
mobile devices when TKIP and WPA’s MIC protocols are implemented. We
observed in the results that decrease on performance was more noticeable on
PDAs than other devices such as laptops
1 Introduction
Ever since wireless networks appeared, many questions concerning security issues
were made. WEP (Wired Equivalent Privacy) was part of IEEE’s 802.11 standard,
and it attempted to provide secure wireless communications.
In 802.11 WEP uses a secret 40 bit key (weak) or 128 bit key (strong) in 802.11b and
a pseudorandom
number generator (RC4). Two processes are applied to clear text;
one of them ciphers data and the other one protects it from unauthorized modifica-
tions while in transit. The secret key is concatenated with a random initialization
vector (IV) that adds 24 bits to the resulting key. This key is processed in the pseudo-
random number generator that outputs a large pseudorandom key stream. The trans-
mitter combines it with the clear text using an XOR operation, creates the ciphered
text and sends it to the receiver along with the IV. When the receiver gets the ci-
phered text, it uses the IV and its own copy of the secret key to generate the same key
stream as the transmitter. The receiver combines them with the XOR operation and
generates the original clear text.
Ochoa Indart A. and Arturo Pérez Díaz J. (2004).
ANALYSIS OF WEP PERFORMANCE ON MOBILE DEVICES.
In Proceedings of the 2nd International Workshop on Security in Information Systems, pages 324-330
DOI: 10.5220/0002686203240330
Copyright
c
SciTePress
In order to protect the ciphered text against modifications while it is in transit, WEP
applies an integrity checking algorithm (CRC-32) to the clear text and generates an
integrity check value (ICV).
The ICV is concatenated to the text before it is encrypted with the key and is sent to
the receptor along with the IV. When the checking algorithm is applied to the clear
text and is compared with the output with the ICV value received, it can be verified if
there has been any modification. [1]
However as Nikita Borisov et. al demonstrated, the WEP checksum is a linear func-
tion of the message. One consequence of the above property is that it becomes possi-
ble to make controlled modifications to a ciphertext without disrupting the checksum.
[2].
Description Processor RAM WLAN NIC OS
Laptop Client 1 – HP
ze5785 us
Intel Pentium
4 2.4 Ghz.
512 MB LAN-Express
IEEE 802.11b
NIC
Windows XP
Home Edi-
tion
Laptop Client 2 – IBM
Think Pad 2655
Intel Pentium
3 1 Ghz.
128 MB Proxim IEEE
802.11 b/g PC
Card.
Windows
2000 Profes-
sional
PDA Client – HP iPAQ
4155
Intel XScale
400 Mhz.
64 MB Embedded Windows
Mobile 2003
Server Laptop – HP
ze5385 us
Intel Pentium
4 2.66 Ghz.
512 MB LAN-Express
IEEE 802.11b
NIC
Windows XP
Home Edi-
tion
WEP uses the RC4 symmetric stream cipher for encryption and decryption purposes.
Symmetric means that the sender and receiver must use the same key for proper en-
cryption and decryption functions. [3]
There are other key lengths for WEP, such as 64 bits, which was used in our tests.
There are various types of known attacks against WEP, and it is not considered se-
cure. Although there are other ciphering techniques, WEP is implemented natively in
many OS such as Windows XP, Windows Mobile and Palm OS. This is why WEP is
still widely used.
Design of secure protocols is difficult, and fraught with many complications. It re-
quires special expertise beyond that acquired in engineering network protocols. A
good understanding of cryptographic primitives and their properties is critical. From a
purely engineering perspective, the use of CRC-32 and RC4 can be justified by their
speed and ease of implementation. [2]
Mobile devices such as PDA’s are being increasingly used in Wireless LANs
(WLANs); these devices have limited processing resources; and therefore, the impact
on data transfer performance is of particular interest because of the processing over-
head it causes.
325
There are other security protocols such as PEAP or LEAP, which promise better
protection, however, it has been proofed that there are other attacks that could affect
them such as the ones published by Mishra and Arbaugh, which explains that 802.11
frames, including 802.1X messages, are easily sniffed. For this reason, IEEE 802.11
Task Group I recommends EAP methods resistant to dictionary attack.
It's worth heeding this advice, since dictionary attacks enable an attacker to recover
the user password, which often can provide access to more than just the 802.11 net-
work. Therefore these attacks are more serious than the previously documented WEP
attacks and customers using 802.1X should strongly consider adopting dictionary
attack-resistant authentication methods such as EAP TLS, SRP, TTLS and PEAP. [4]
LEAP is a type of Radius EAP. It is used to authenticate access by a wireless client
(typically a laptop or pc) to a wireless router, typically a Cisco Aironet base sta-
tion.[5]
RADIUS is a widely deployed protocol for network access authentication, authoriza-
tion and accounting (AAA). [6]
This paper presents an analysis of the data transfer performance achieved by laptops
and PDA’s when using 64 and 128 bit keys with WEP and when transmitting clear
text using an infrastructure WLAN.
2 Experimental Section
2.1 Equipment Used
Two laptops and a PDA were used as clients. A third laptop was used as server. A
brief description of the equipment can be found in table 1.
The access point that was used was a Microsoft Broadband Networking Wireless
Base Station Model MN-500, which is Wi-Fi certified.
2.2 Performance measurement
In order to obtain performance measurements of common uses of a WLAN, a simple
web-based script was written in PHP, running on an Apache 2.0.48 web server with
PHP Engine 4.0.1. Measurements were stored using mySQL 4.0.13.
The PHP script sends a random stream of bytes, ranging from 100 to 5000 kilobytes.
Three fields are stored in the database, the client’s IP address, the amount of data
transferred and the time that the transfer took.
326
The resulting web page is reloaded 5 seconds after the transfer is finished and a new
stream of different size is sent to the client.
2.3 Test scenarios
Several tests were performed, in order to test different situations and compare them.
The first variable is the length of the key, three different scenarios were tested in this
case, with no key (no WEP encryption), 64 bit, and 128 bit keys.
The second variable is distance, 3 different distances were tested. In every case, all
the devices were at the same distance.
a) Five feet away from the Access Point. No interferences.
b) Twelve feet away from the Access Point. No interferences.
c) Forty feet away from the Access Point. On the second floor, home environment
(Computers and PDA were on the first floor).
For each scenario, 1200 samples were gathered, 400 for every mobile device.
Using the gathered data, simple statistical analysis was calculated, specifically, the
mean value of the samples and the standard deviation.
3 Results and Discussion
3.1 Performance with no WEP encryption
5ft 12ft 40ft
HP Laptop 170.21 148.9 122.83
IBM Laptop 169.26 145.8 120.46
iPAQ PDA 168.6 148.39 119.14
Table 2: Average results in KB/S with no WEP encryption.
327
Figure 1: Data transfer
performance no WEP
encryption
0
50
100
150
200
5ft 12ft 40ft
KB/ S
HP Laptop
IBM Laptop
iPAQ PDA
When using no WEP encryption, the performance loss is similar on both laptops and
the PDA as shown in Table 2 and Figure 1.
3.2 Performance with 64 bit key WEP encryption
5ft 12ft 40ft
HP Laptop 162.19 147.32 112.14
IBM Lap-
top
158.22 149.41 115.11
iPAQ PDA 154.63 141.32 104.78
Table 3: Average results in KB/S with WEP and a 64 bit key.
Test results with a 64 bit key show that the PDA’s performance was more noticeable
than both laptops. This can be observed in Table 3 and Figure 2.
Figure 2: Data transfer
performance 64 bit key WEP
encryption
0
50
100
150
200
5ft 12ft 40ft
KB/S
HP Laptop IBM Laptop
iPAQ PDA
328
3.3 Performance with 128 bit key WEP encryption
5ft 12ft 40ft
HP Laptop 147.24 140.33 118.75
IBM Lap-
top
150.81 145.28 117.63
iPAQ PDA 140.29 134.5 90.69
Table 3: Average results in KB/S with WEP and a 128 bit key.
It is clear that the PDA decreased its performance more than laptops. This can be seen
in Table 3 and figure 4.
Figure 3: Data transfer
performance 128 bit key WEP
encryption
0
50
100
150
200
5ft 12ft 40ft
KB/S
HP Laptop IBM Laptop
iPAQ PDA
3.4 Results Analysis
No WEP 64 bit 128 bit
HP Laptop 147.31 140.55 135.44
IBM Lap-
top
145.17 140.91 137.90
iPAQ PDA 145.37 133.57 121.82
Table 4: Overall Performance in KB/S
It is clearly visible that the PDA’s performance (See Table 4) was considerably re-
duced by WEP encryption. It is clear that the reduced computing power of the PDA
resulted in a bigger impact on performance.
As mentioned above, WEP uses symmetric keys, because of that, we expected better
performance results on the PDA, but it affected it visibly. We would now expect that
using EAP-TLS or other similar technique the performance loss to be greater.
TKIP changes the ciphering key very often, and requires much more resources. Based
on this, we can extrapolate the results and consider that when using TKIP, the per-
formance loss will be much bigger.
329
Both laptops had similar behavior, and they were not visibly affected by WEP en-
cryption.
We can see an overall comparison of performance in figure 4.
Figure 4: Overall performance
comparison
0
50
100
150
200
No WEP 64 bit 128 bit
KB/S
HP Laptop IBM Laptop
iPAQ PDA
4 Future Works
We will repeat these tests with ciphering techniques specified by WPA and evaluate
their performance in order to search alternatives for mobile devices if there is a con-
siderable loss of performance.
5 Conclusions
Approximately, the PDA lowered its performance to 83.80% compared to the 91.94%
observed in Client 1 and 94.99 % of Client 2, when looking their performance based
on no WEP encryption and 128 bit encryption.
From the standard deviations observed, the PDA had the lowest levels overall, this
can be because laptops usually run other processes on the background that might
impact some measurements.
Security is vital to wireless communications, there has been a big amount of effort
and research to provide reliable ciphering techniques. Progress has been achieved in
this field; however there are new scenarios where wireless communications were not
very popular a few years ago.
Mobile devices have limited resources and processing power, this is why, ciphering
techniques used in these devices, have to meet their constraints and yet meet security
levels.
It will be vital to take these constraints when designing new security schemes, and
when these schemes are deployed to new operating systems for mobile devices, they
must allow limited devices to work properly, without degrading QoS and providing
secure, reliable data transfers.
330
WPA security protocols are expected to consume more resources than old protocols
such as WEP, so special protocols for limited devices should be developed, so their
performance is not affected.
References
[1] Nichols, Randall and Lekkas, Panos. Wireless Security: Models, Threats, and Solutions.
McGraw Hill. Edition 1, 2002. ISBN: 0071380388
[2] Nikita Borisov, Ian Goldberg, David Wagner. Intercepting Mobile Communications: The
Insecurity of 802.11
http://www.isaac.cs.berkeley.edu/isaac/wep-draft.pdf
[3] Shon Harris, Latest trends in wireless security
http://a907.g.akamai.net/7/907/3644/v0001/ntschool1.download.akamai.com/3644/_vl/Arti
cles/WIRELESS-SECURITY-SHONa.pdf
[4] Arunesh Mishra and William A. Arbaugh, An Initial Security Analysis of the IEEE 802.1X
Standard, University of Maryland.
http://www.cs.umd.edu/~waa/1x.pdf
[5] Cameron MacNally, Cisco LEAP protocol description
http://www.missl.cs.umd.edu/wireless/ethereal/leap.txt
[6] Bernard Aboba. Wireless LAN Security Site. http://www.drizzle.com/~aboba/IEEE/
331